In brief: A faster task does not, on its own, show the return on AI. Understanding that return means following what changes in the work and how those changes affect the business. The AIG Insurance results reported in Anthropic’s S-1 identify operational changes, not a complete ROI calculation. The business case is built on what happened after the task became faster.
An enterprise can spend more on AI while still learning where it creates value. To understand the return, we need to follow what changes in the work and how those changes affect the business.
What Anthropic’s S-1 reports about AIG
Anthropic’s October 26 S-1 includes a case study on AIG Insurance. It reports that Lexington’s Middle Market Property business increased quoted submissions by 30%, reduced time to quote by 55%, and increased submissions bound by 40%, as of March 31, 2026. These company/vendor-reported figures have not been independently verified. They describe quoted submissions, time to quote, and submissions bound. Source
Those results were already captured in the AIDE Institute’s use-case catalogue when we compared the records with the filing.
Evaluating AI in your organization? Talk with the AIDE Institute about research and data to inform your assessment.
What happened to the additional capacity?
A shorter time to quote can allow underwriters to consider more submissions. It could also improve responsiveness to brokers or free time for difficult cases. Each possibility has a different economic consequence.
If the team handles more business with the same resources, the benefit may appear in growth relative to costs. If people spend the freed time on judgment and service, the benefit may appear in quality or customer outcomes. Cash savings or avoided costs need their own baseline and evidence.
These are possible pathways from faster work to business value. The reported time reduction alone does not tell us which occurred or how much value it produced.
More business also needs an economic explanation
An increase in submissions bound brings the analysis closer to commercial results. In insurance, however, more business is valuable only in the context of the risk accepted, pricing, expected losses, and the costs of serving it.
A complete evaluation would ask whether the additional business met the insurer’s profitability and risk objectives, and how much of the change can reasonably be attributed to AI rather than market conditions, pricing decisions, or other operational changes.
For an enterprise considering a similar investment, we would ask:
- What changed in a defined workflow, over what period, and compared with which baseline?
- What did the organization do with the capacity it gained?
- Did revenue, costs, quality, service, or risk change as a result?
- What did implementation and ongoing operation cost, including integration, review, training, and maintenance?
Studying the enterprise
The AIDE Institute’s research examines public evidence of AI adoption alongside documented applications and reported results. These provide different views of how a company is changing.
An application can show where AI is being used. An operational result can show what changed in that work. Understanding whether the enterprise benefits requires looking across workflows, organizational decisions, costs, and financial performance without counting the same benefit twice.
The AIG figures do not provide a complete ROI calculation. They identify operational changes whose economic consequences can be investigated.
For leaders evaluating AI, the next question is what happened after the task became faster. That is where the business case needs to be built.
Sources
Editorial credit: Independently edited by the AIDE Institute.
